Regio- and Enantioselective N-Allylations of Imidazole, Benzimidazole

May 29, 2009 - 5, 2a (4), 3d (3-Me-C6H4), 5, 21d, 89:11, 75, 95:5, 90. 6, 2a (4) ...... Lee , J., Doucette , A., Wilson , N. S., and Lord , J. Tetrahe...
0 downloads 0 Views 922KB Size
Download PDF
Published on Web 05/29/2009

Regio- and Enantioselective N-Allylations of Imidazole, Benzimidazole, and Purine Heterocycles Catalyzed by Single-Component Metallacyclic Iridium Complexes Levi M. Stanley and John F. Hartwig* Department of Chemistry, UniVersity of Illinois, 600 South Mathews AVenue, Urbana, Illinois 61801 Received March 30, 2009; E-mail: [email protected]

Abstract: Highly regio- and enantioselective iridium-catalyzed N-allylations of benzimidazoles, imidazoles, and purines have been developed. N-Allylated benzimidazoles and imidazoles were isolated in high yields (up to 97%) with high branched-to-linear selectivity (up to 99:1) and enantioselectivity (up to 98% ee) from the reactions of benzimidazole and imidazole nucleophiles with unsymmetrical allylic carbonates in the presence of single component, ethylene-bound, metallacyclic iridium catalysts. N-Allylated purines were also obtained in high yields (up to 91%) with high N9/N7 selectivity (up to 96:4), high branched-to-linear selectivity (98:2), and high enantioselectivity (up to 98% ee) under similar conditions. The reactions encompass a range of benzimidazole, imidazole, and purine nucleophiles, as well as a variety of unsymmetrical aryl, heteroaryl, and aliphatic allylic carbonates. Competition experiments between common amine nucleophiles and the heterocyclic nitrogen nucleophiles studied in this work illustrate the effect of nucleophile pKa on the rate of iridium-catalyzed N-allylation reactions. Kinetic studies on the allylation of benzimidazole catalyzed by metallacyclic iridium-phosphoramidite complexes, in combination with studies on the deactivation of these catalysts in the presence of heterocyclic nucleophiles, provide insight into the effects of the structures of the phosphoramidite ligands on the stability of the metallacyclic catalysts. The data obtained from these studies have led to the development of N-allylations of benzimidazoles and imidazoles in the absence of an exogenous base.

Introduction

Enantioselective, metal-catalyzed allylic amination is a useful method to prepare enantiomerically enriched allylic amines from a variety of nitrogen nucleophiles and allylic electrophiles.1-10 However, metal-catalyzed enantioselective N-allylation of less nucleophilic reagents, such as nitrogen-containing heterocycles, remains underdeveloped, despite the synthetic utility of the products.11,12 Currently, the most prominent catalysts for enantioselective N-allylation of nitrogen-containing heterocycles are chiral, nonracemic palladium complexes. A report by Trost and co-workers on the palladium-catalyzed enantioselective N-allylation of a guanine equivalent with a cyclic meso-diester as a key step in the total synthesis of the carbocyclic nucleoside (1) Lu, Z.; Ma, S. Angew. Chem., Int. Ed. 2008, 47, 258–297. (2) Graening, T.; Schmalz, H.-G. Angew. Chem., Int. Ed. 2003, 42, 2580– 2584. (3) Trost, B. M. J. Org. Chem. 2004, 69, 5813–5837. (4) Trost, B. M.; Crawley, M. L. Chem. ReV. 2003, 103, 2921–2944. (5) Dai, L.-X.; Tu, T.; You, S.-L.; Deng, W.-P.; Hou, X.-L. Acc. Chem. Res. 2003, 36, 659–667. (6) Agrofoglio, L. A.; Gillaizeau, I.; Saito, Y. Chem. ReV. 2003, 103, 1875–1916. (7) Trost, B. M. Acc. Chem. Res. 1996, 29, 355–364. (8) Trost, B. M.; Van Vranken, D. L. Chem. ReV. 1996, 96, 395–422. (9) Helmchen, G.; Dahnz, A.; Dubon, P.; Schelwies, M.; Weinhofen, R. Chem. Commun. 2007, 675–691. (10) Takeuchi, R.; Kezuka, S. Synthesis 2006, 3349–3366. (11) Pozharskii, A. F.; Soldatenkov, A. T.; Katritzky, A. R. Heterocycles in Life and Society; Wiley: New York, 1997. (12) Katritzky, A. R.; Pozharskii, A. F. Handbook of Heterocyclic Chemistry, 2nd ed.; Pergamon: Oxford, 2002. 10.1021/ja902243s CCC: $40.75  2009 American Chemical Society

(-)-carbovir illustrates the utility of enantioselective N-allylations of heterocyclic nucleophiles.13 Additional reports detailing enantioselective palladium-catalyzed N-allylations of heterocyclic nucleophiles have followed, but these processes are limited by the inability to generate chiral allylic substitution products selectively from achiral linear allylic esters.14-16 Over the past decade, enantioselective iridium-catalyzed allylic amination has emerged as a powerful complement to enantioselective, palladium-catalyzed allylic amination reactions.17-40 Iridium-catalyzed allylic amination reactions make possible the synthesis of enantioenriched allylic amines from achiral linear reactants. However, enantioselective reactions that selectively form branched allylic substitution products from nitrogen heterocycles, other than phthalimide, and achiral linear reactants are unknown (eq 1).41 These reactions would create a direct approach to enantioenriched heterocycles that cannot be accessed by hydrogenation of a CdN double bond or classical resolution of racemic material and that can be transformed into a series of further functionalized materials.42 Access to enantiomerically enriched N-allylated heterocycles from unsym(13) Trost, B. M.; Madsen, R.; Guile, S. G.; Elia, A. E. H. Angew. Chem., Int. Ed. Engl. 1996, 35, 1569–1572. (14) Trost, B. M.; Shi, Z. J. Am. Chem. Soc. 1996, 118, 3037–3038. (15) Trost, B. M.; Madsen, R.; Guile, S. D.; Brown, B. J. Am. Chem. Soc. 2000, 122, 5947–5956. (16) Trost, B. M.; Dong, G. J. Am. Chem. Soc. 2006, 128, 6054–6055. (17) Spiess, S.; Welter, C.; Franck, G.; Taquet, J.-P.; Helmchen, G. Angew. Chem., Int. Ed. 2008, 47, 7652–7655. (18) Bondzic, B. P.; Farwick, A.; Liebich, J.; Eilbracht, P. Org. Biomol. Chem. 2008, 6, 3723–3731. J. AM. CHEM. SOC. 2009, 131, 8971–8983

9

8971

ARTICLES

Stanley and Hartwig

metrical allylic electrophiles is presently limited to rhodiumcatalyzed reactions of N3-benzoyl thymine or a dihydropyrimidin2(3H)-one with enantioenriched secondary allylic carbonates with retention of configuration.43,44 The products of these reactions are relevant to medicinal chemistry because of the established biological activity associated with benzimidazole,45 imidazole,46 and purine derivatives.47 For example, the chiral N-allyl imidazoles are advanced intermediates to a class of kinase inhibitors,48,49 while the N-allyl purines are chiral precursors to a family of antiretroviral drugs.50-53 Furthermore, the N-allyl heterocycle products contain a terminal olefin that makes possible the straightforward (19) Pouy, M. J.; Leitner, A.; Weix, D. J.; Ueno, S.; Hartwig, J. F. Org. Lett. 2007, 9, 3949–3952. (20) Markovic, D.; Hartwig, J. F. J. Am. Chem. Soc. 2007, 129, 11680– 11681. (21) Lee, J. H.; Shin, S.; Kang, J.; Lee, S. J. Org. Chem. 2007, 72, 7443– 7446. (22) Spiess, S.; Berthold, C.; Weinhofen, R.; Helmchen, G. Org. Biomol. Chem. 2007, 5, 2357–2360. (23) Yamashita, Y.; Gopalarathnam, A.; Hartwig, J. F. J. Am. Chem. Soc. 2007, 129, 7508–7509. (24) Defieber, C.; Ariger, M. A.; Moriel, P.; Carreira, E. M. Angew. Chem., Int. Ed. 2007, 46, 3139–3143. (25) Singh, O. V.; Han, H. J. Am. Chem. Soc. 2007, 129, 774–775. (26) Nemoto, T.; Sakamoto, T.; Matsumoto, T.; Hamada, Y. Tetrahedron Lett. 2006, 47, 8737–8740. (27) Shekhar, S.; Trantow, B.; Leitner, A.; Hartwig, J. F. J. Am. Chem. Soc. 2006, 128, 11770–11771. (28) Polet, D.; Alexakis, A.; Tissot-Croset, K.; Corminboeuf, C.; Ditrich, K. Chem.sEur. J. 2006, 12, 3596–3609. (29) Leitner, A.; Shekhar, S.; Pouy, M. J.; Hartwig, J. F. J. Am. Chem. Soc. 2005, 127, 15506–15514. (30) Welter, C.; Moreno, R. M.; Streiff, S.; Helmchen, G. Org. Biomol. Chem. 2005, 3, 3266–3268. (31) Weihofen, R.; Dahnz, A.; Tverskoy, O.; Helmchen, G. Chem. Commun. 2005, 3541–3543. (32) Polet, D.; Alexakis, A. Org. Lett. 2005, 7, 1621–1624. (33) Leitner, A.; Shu, C.; Hartwig, J. F. Org. Lett. 2005, 7, 1093–1096. (34) Miyabe, H.; Matsumura, A.; Moriyama, K.; Takemoto, Y. Org. Lett. 2004, 6, 4631–4634. (35) Shu, C.; Leitner, A.; Hartwig, J. F. Angew. Chem., Int. Ed. 2004, 43, 4797–4800. (36) Tissot-Croset, K.; Polet, D.; Alexakis, A. Angew. Chem., Int. Ed. 2004, 43, 2426–2428. (37) Welter, C.; Koch, O.; Lipowsky, G.; Helmchen, G. Chem. Commun. 2004, 896–897. (38) Kiener, C. A.; Shu, C.; Incarvito, C.; Hartwig, J. F. J. Am. Chem. Soc. 2003, 125, 14272–14273. (39) Lipowsky, G.; Helmchen, G. Chem. Commun. 2004, 116–117. (40) Ohmura, T.; Hartwig, J. F. J. Am. Chem. Soc. 2002, 124, 15164– 15165. (41) Weihofen, R.; Tverskoy, O.; Helmchen, G. Angew. Chem., Int. Ed. 2006, 45, 5546–5549. (42) Gandelman, M.; Jacobsen, E. N. Angew. Chem., Int. Ed. 2005, 44, 2393–2397. (43) Evans, P. A.; Lai, K. W.; Zhang, H.-R.; Huffman, J. C. Chem. Commun. 2006, 844–846. (44) Evans, P. A.; Qin, J.; Robinson, J. E.; Bazin, B. Angew. Chem., Int. Ed. 2007, 46, 7417–7419. (45) Alamgir, M.; Black, D. S. C.; Kumar, N. Top. Heterocycl. Chem. 2007, 9, 87–118. (46) De Luca, L. Curr. Med. Chem. 2006, 13, 1–23. (47) Legraverend, M.; Grierson, D. S. Bioorg. Med. Chem. 2006, 14, 3987– 4006. (48) Graczyk, P. P.; et al. Bioorg. Med. Chem. Lett. 2005, 15, 4666–4670. (49) Rech, J. C.; Yato, M.; Duckett, D.; Ember, B.; LoGrasso, P. V.; Bergman, R. G.; Ellman, J. A. J. Am. Chem. Soc. 2007, 129, 490– 491. (50) Simons, C. Nucleoside Mimetics. Their Chemistry and Biological Properties.; Gordon and Breach: Australia, 2001. (51) Song, G. Y.; Paul, V.; Choo, H.; Morrey, J.; Sidwell, R. W.; Schinazi, R. F.; Chu, C. K. J. Med. Chem. 2001, 44, 3985–3993. (52) Agrofoglio, L. A.; Challand, S. R. Acyclic, Carbocyclic and LNucleosides; Kluwer Academic: Dordrecht, 1998. (53) Anastasi, C.; Quelever, G.; Burlet, S.; Garino, C.; Souard, F.; Kraus, J.-L. Curr. Med. Chem. 2003, 10, 1825–1843. 8972

J. AM. CHEM. SOC.

9

VOL. 131, NO. 25, 2009

Figure 1. Cyclometalated iridium catalyst precursors (1, 2a, and 2b) and

phosphoramidite ligands (L1 and L2).

syntheses of R-imidazolyl alcohol, R-purinyl alcohol, β-imidazolyl alcohol, and β-purinyl alcohol derivatives.

We report the formation of enantioenriched, branched N-allyl heterocycles from reactions of benzimidazoles, imidazoles, and purines with achiral linear allylic carbonates in the presence of single-component iridium catalysts. The new iridium catalysts, along with the proper base, improve the efficiency of the allylation process, overcome isomerization of the product to the enamine, and lead to the formation of products with high enantioselectivity, high regioselectivity for addition to the more substituted position of the allyl electrophile, and high selectivity for the addition to the N9 position over the N7 position of purines. A combination of competition experiments, kinetic studies, and experiments on catalyst deactivation create an improved understanding of the activity and stability of the metallacyclic iridium catalysts. In addition to revealing a new class of asymmetric N-allylation reactions, these studies led to the identification of an appropriate iridium catalyst for basefree N-allylations of benzimidazole and imidazole nucleophiles. Results and Disscussion Catalyst Identification and Optimization of Reaction Conditions. We previously showed that complex 1 (Figure 1), which

is generated from [Ir(COD)Cl]2, phosphoramidite ligand L1 in Figure 1, and base, catalyzes asymmetric allylic substitutions with weakly basic nitrogen nucleophiles (arylamines), as well as basic nitrogen nucleophiles (benzylic amines and alkylamines), to form branched allylic substitution products.38 Thus, initial studies to develop iridium-catalyzed asymmetric Nallylations of nitrogen heterocycles containing acidic N-H bonds were conducted with preformed metallacycle 1 as catalyst. The model reaction of methyl cinnamyl carbonate (3a) with benzimidazole (4a) was selected to test the viability of metallacycle 1 as a catalyst for allylic substitutions with nitrogencontaining heterocycles. In contrast to reactions of amines with methyl cinnamyl carbonate, no reaction occurred between benzimidazole and methyl cinnamyl carbonate (Table 1, entry 1). The same reaction with added base did occur, but this reaction conducted without

Ir-Catalyzed N-Allylation of Heterocycles

ARTICLES

Table 1. Study of Metallacyclic Iridum Catalyst Precursors and Reaction Conditions for the Allylation of Benzimidazole with Methyl Cinnamyl Carbonatea

entry

1 2 3 4f,g 5 6g,i 7 8 9 10 11 12 13 14 15

catalyst (mol %)

1 (2) 1 (2) 1 (2) 1 (2) 1 (1) 1 (1) 2a (2) 2b (2) 2a (2) 2a (2) 2a (2) 2a (2) 2a (2) 2a (2) 2a (2)

additive (mol %)

[Ir(COD)Cl]2 (1) [Ir(COD)Cl]2 (1) [Ir(COD)Cl]2 (1) [Ir(COD)Cl]2 (1) [Ir(COD)Cl]2 (0.5) [Ir(COD)Cl]2 (0.5) ----------

base

-Cs2CO3 K3PO4 K3PO4 K3PO4 K3PO4 K3PO4 K3PO4 Na3PO4 Li3PO4 KOt-Bu NaOt-Bu Cs2CO3 K2CO3 Na2CO3

time (h)

conversion (%)b

5a/6a/7ab

yield (%)c

10 4 4 4 16 24 2 2 3 3 3 3 3 3 3

93:7, the N9 branched-to-linear ratios were >90:10, and the N9 substitution products 21 were isolated in >75% yield with >90% ee. One exception to these trends was the reaction of 6-chloropurine with methyl ortho-methoxycinnamyl carbonate 3f, which formed the N-allylated purine product 21f with moderate enantioselectivity in the presence of either catalyst 2a or catalyst 2b (entries 7 and 8, 2af78% ee, 2bf81% ee). The reaction of 6-chloropurine with aliphatic carbonate 3k in the presence of 4 mol % 2a occurred with poor N9/N7 selectivity (74:26, entry 9). However, the same reaction conducted with 4 mol % 2b as the catalyst occurred with high regio- and enantioselectivity (N9/N7 ) 93:7, entry 10). The difference in N9/N7 selectivity for the reaction of 6-chloropurine with 3k in the presence of parent catalyst 2a and ortho-OMe catalyst 2b is probably due to a faster decomposition of catalyst 2a than of catalyst 2b. The relative rates of catalyst decomposition are discussed in greater detail later in this paper. Several other purines underwent the allylation process in high yields and with high selectivities (Table 6). For example, reactions of 6-(methylthio)purine 20b and adenine 20c with methyl cinnamyl carbonate occurred to form products 21h and 21i with high N9/N7 and N9 branched-to-linear selectivities (entries 1 and 2). From these reactions, the branched allylated purine derivatives 21h and 21i were isolated in high yields with high enantioselectivities. The reaction of adenine conducted with a lower 2 mol % loading of parent catalyst 2a occurred with a similar yield and selectivity for formation of 21i as the reaction conducted with 4 mol % 2a (entry 3). In contrast, the allylations of bis-Boc-adenine 20d and 2-amino-6-chloropurine 20e, a guanine equivalent, with methyl cinnamyl carbonate occurred with higher selectivities when conducted with ortho-OMe catalyst 2b than with the parent catalyst 2a. The reaction of bis-Boc-adenine 20d with methyl cinnamyl carbonate to form product 21j proceeded to low conversion (entry 4) in the presence of 2 mol % 2a but occurred to high conversion with (72) The preference for N9-selectivity was confirmed by NMR spectroscopy (HMBC) for compound 21e. See the Supporting Information for additional details. J. AM. CHEM. SOC.

9

VOL. 131, NO. 25, 2009

8977

ARTICLES

Stanley and Hartwig

Scheme 3. Direct Synthesis of 18 by Iridium-Catalyzed Allylic Substitution as a Formal Synthesis of JNK3 Inhibitor 16

Scheme 4. Potential Products from an Iridium-Catalyzed Allylation of Purine Nucleophiles

high N9/N7 selectivity, branched-to-linear selectivity, and enantioselectivity when conducted in the presence of 2 mol % 2b (entry 5). Likewise, the reaction of 2-amino-6-chloropurine 20e with methyl cinnamyl carbonate occurred with modest N9/ N7 selectivity (82:18, entry 6) in the presence of 2 mol % 2a but occurred with excellent N9/N7 selectivity when conducted in the presence of 2 mol % 2b (entry 7). Application of the Enantioselective N-Allylation of Purine Nucleophiles. (9H-Purin-9-yl)alcohols are key substructures of

the commercial antivirals adefovir and tenofovir, which are adenine analogues (Figure 4). Adefovir is FDA approved for treatment of hepatitis B infections, while tenofovir is FDA approved for treatment of HIV and hepatitis B infections.73-76 Analogues of current antiviral compounds are important for addressing the spread of viral resistance. Thus, we sought to illustrate how chiral N-allylated adenine derivatives could be converted to R-adeninyl alcohols and β-adeninyl alcohols, which are optically active R-substituted analogues of the antiretrovirals adefovir and tenofovir (Scheme 5). Ozonolysis of bis-BOCprotected adenine derivatives 21j and 21l, followed by reduction of the resulting ozonides with NaBH4, formed bis-Boc-protected (S)-2-(6-amino-9H-purin-9-yl)-2-phenylethanol 25a and bisBoc-protected (S)-2-(6-amino-9H-purin-9-yl)pentan-1-ol 25b in 89% and 88% yield, respectively, without erosion of enantiomeric purity (eq 6). Standard procedures for hydroboration and oxidation gave complex product mixtures from reactions of bisBOC-protected adenine derivatives 21j and 21l.77,78 However, hydroboration of the monomethoxytrityl(MMTr)-protected ad(73) De Clercq, E. Med. Res. ReV. 2008, 28, 929–953. (74) De Clercq, E. Future Virol. 2006, 1, 709–715. (75) Cihlar, T.; Delaney, W. E., IV.; Mackman, R. Modified Nucleosides 2008, 601–630. (76) Tillmann, H. L. Therapeut. Clin. Risk. Manag. 2008, 4, 797–802. (77) Brown, H. C.; Chen, J. J. Org. Chem. 1981, 46, 3978–3988. (78) Kabalka, G. W.; Shoup, T. M.; Goudgaon, N. M. J. Org. Chem. 1989, 54, 5930–5933. 8978

J. AM. CHEM. SOC.

9

VOL. 131, NO. 25, 2009

enine derivatives 26a and 26b, followed by oxidation of the resulting alkylboranes in the presence of basic hydrogen peroxide, gave MMTr-protected (R)-3-(6-amino-9H-purin-9-yl)3-phenylpropan-1-ol) 27a and MMTr-protected (S)-3-(6-amino9H-purin-9-yl)hexan-1-ol 27b in 82% and 77% yield (eq 7). Thus, iridium-catalyzed allylation, followed by a combination of ozonolysis and reduction or a combination of hydroboration and oxidation, provides a new route to chiral, nonracemic derivatives of adenine that are related to FDA approved reverse transcriptase inhibitors. Competition between Azole and Amine Nucleophiles. After developing methods for the regio- and enantioselective Nallylation of benzimidazoles, imidazoles, and purines, we conducted a series of competition experiments to develop an understanding of the relative nucleophilicities of these heterocycles and other common nitrogen nucleophiles toward iridiumcatalyzed allylic substitution reactions (Scheme 6). These experiments were designed to provide a basis to rationalize the high yields for reactions of the imidazole, benzimidazole, and purine reagents, which are typically weaker nucleophiles than amines, toward metal-catalyzed allylic substitution reactions. Mayr’s scale of nucleophilicity for nitrogen nucleophiles provides benchmark reactivities of benzylamine, aniline, and imidazole with S-methyldibenzothiophenium ion and a variety of benzhydrylium ions as electrophiles.79,80 The nucleophilicity and nucleophile-specific slope parameters for these nucleophiles show that benzylamine reacts twice as fast as aniline and 20 times faster than imidazole with the activated electrophiles studied by Mayr and co-workers. We conducted competition experiments between imidazole and benzylamine, aniline, benzimidazole, and bis-BOC-adenine to determine whether this trend in nucleophilicity, based on reactions of nitrogen nucleophiles with S-methyldibenzothiophenium ion and benzhydrylium ions as electrophiles, holds for iridium-catalyzed N-allylation reactions. Reactions of imidazole (2 equiv) and either benzylamine, aniline, benzimidazole, or bis-Boc-adenine (2 equiv) with methyl cinnamyl carbonate 3a (1.0 equiv) were performed in the presence of K3PO4 (1.0 equiv) and 4 mol % 2b. The reaction of 3a with benzylamine and imidazole gave the N-allylamine and imidazole products 28 and 11a in a 74:26 ratio (eq 8). Thus, benzylamine reacts faster than imidazole in this iridiumcatalyzed process, as expected, but the 3:1 ratio of relative rates was much smaller than the 20:1 ratio that would be expected based on Mayr’s nucleophile parameters for benzyl amine (N ) 13.46, s ) 0.62) and imidazole (N ) 10.41, s ) 0.70). The discrepancy is best explained by a contribution from the reaction (79) Phan, T. B.; Breugst, M.; Mayr, H. Angew. Chem., Int. Ed. 2006, 45, 3869–3874. (80) Brotzel, F.; Chu, Y. C.; Mayr, H. J. Org. Chem. 2007, 72, 3679– 3688.

Ir-Catalyzed N-Allylation of Heterocycles

ARTICLES

Table 5. Iridium-Catalyzed Allylation of 6-Chloropurinea

entry

catalyst (mol %)

3 (R)

time (h)

21

N9/N7b,c

yield (%)d

21/22b

ee (%)e

1 2 3 4 5 6 7 8 9 10

2a (2) 2a (4) 2a (4) 2a (4) 2a (4) 2a (4) 2a (4) 2b (4) 2a (4) 2b (4)

3a (Ph) 3a (Ph) 3b (4-MeO-C6H4) 3c (4-Br-C6H4) 3d (3-Me-C6H4) 3e (2-Naphthyl) 3f (2-MeO-C6H4) 3f (2-MeO-C6H4) 3k (Heptyl) 3k (Heptyl)

8 5 5 5 5 5 4 4 5 2

21a 21a 21b 21c 21d 21e 21f 21f 21g 21g

94:6 94:6 96:4 94:6 89:11 93:7 95:5 94:6 74:26 93:7

63 83 91 76 75 77 60 84 68 85

93:7 93:7 98:2 91:9 95:5 93:7 95:5 90:10 92:8 91:9

94 93 92 95 90 92 78 81 96 94

a See Supporting Information for experimental details. b Determined by 1H NMR spectroscopy of the crude reaction mixture. c N9/N7 ratio indicates the ratio of N9 allylation to N7 allylation (21+22)/(23+24). d Isolated yield of 21. e Determined by chiral HPLC.

Table 6. Allylation of Various Purine Nucleophiles with Methyl Cinnamyl Carbonatea

entry

1 2 3 4 5 6 7

catalyst (mol %)

20

2a (4) 2a (4) 2a (2) 2a (2) 2b (2) 2a (2) 2b (2)

20b (R ) SMe, R ) H) 20c (R1 ) NH2, R2 ) H) 20c (R1 ) NH2, R2 ) H) 20d (R1 ) N(Boc)2, R2 ) H) 20d (R1 ) N(Boc)2, R2 ) H) 20e (R1 ) Cl, R2 ) NH2) 20e (R1 ) Cl, R2 ) NH2) 1

time (h)

21

N9/N7b,c

yield (%)d

21/22b

ee (%)e

5 5 5 6 6 8 4

21h 21i 21i 21j 21j 21k 21k

96:4 96:4 95:5 90:10 92:8 82:18 96:4

91 81 88 29 82 75 83

98:2 93:7 94:6 92:8 92:8 95:5 94:6

92 96 96 93 96 97 98

2

a See Supporting Information for experimental details. b Determined by 1H NMR spectroscopy of the crude reaction mixture. c N9/N7 ratio indicates the ratio of N9 allylation to N7 allylation (21+22)/(23+24). d Isolated yield of 21. e Determined by chiral HPLC.

Scheme 5. Synthesis of New Chiral Derivatives of Adenine Analogue Antivirals

Figure 4. Adenine derivatives that are reverse transcriptase inhibitors.

of imidazolate, rather than imidazole. The imidazolate would be generated by deprotonation of the heterocycle by K3PO4 or the counterion of the iridium-allyl intermediate, which could be the methyl carbonate or methoxide after decarboxylation of the carbonate. If so, then the observed selectivity would result from a competition between benzylamine and the imidazolate or, more precisely, between benzylamine and an equilibrium mixture of the neutral imidazole and the anionic imidazolate. The competition experiment between aniline and imidazole provides further evidence to support this hypothesis. N-Allyl aniline and imidazole products 29 and 11a were formed in a 46:54 ratio favoring the imidazole product 11a (eq 9). This ratio is, again, much smaller than the 10:1 ratio of rates for aniline versus imidazole that would be expected from the nucleophile parameters for these two nucleophiles (N ) 12.64, s ) 0.68

for aniline and N ) 10.41, s ) 0.70 for imidazole). Furthermore, competition experiments between imidazole and either benzimidazole or bis-Boc-adenine favor the formation of the benzimidazole product 5a (eq 10, 11a/5a ) 29:71) or the bisBoc-adenine product 21j (eq 11, 11a/21j ) 15:85), resulting from allylation of the more acidic of the two nucleophiles in J. AM. CHEM. SOC.

9

VOL. 131, NO. 25, 2009

8979

ARTICLES

Stanley and Hartwig

Scheme 6. Competition Experiments between Imidazole and Amine or Heterocyclic Nitrogen Nucleophiles

each case. The combination of these results suggests that imidazole, benzimidazole, and adenine nucleophiles undergo facile iridium-catalyzed N-allylation reactions because of the contributions from reaction of the deprotonated species. The competition experiments also show that an increased acidity of the heterocycle, rather than increased basicity, leads to faster rates for allylation of these heterocycles (10a: pKa ) 18.6; 5a: pKa ) 16.4; 20d: pKa ≈ 15 in DMSO).81 The results from the competition experiments displayed in Scheme 6 suggest that imidazolate, benzimidazolate, and adeninate anions are the form of the heterocycles that undergo the N-allylation reactions. To test whether the methyl carbonate or methoxide anion formed upon generation of the intermediate allyliridium complex or exogenous K3PO4 acted as the base primarily responsible for the formation of the deprotonated (81) Bordwell, F. G. Acc. Chem. Res. 1988, 21, 456–463. 8980

J. AM. CHEM. SOC.

9

VOL. 131, NO. 25, 2009

Scheme 7. Allylation of Benzimidazole with 3a in the Absence of

Exogenous Base

heterocycle species, we studied the reaction of benzimidazole with methyl cinnamyl carbonate catalyzed by 4 mol % 2a or 2 mol % 2b in the absence of K3PO4 at room temperature and at 50 °C (Scheme 7). Two sets of data lead to several conclusions about the relative importance of the potassium phosphate and the counterions of the allyl complex as base. First, the reaction of 3a with 4a in the presence of 4 mol % of the parent catalyst 2a was 68% complete after 5 h at room temperature, while the same reaction conducted at 50 °C was only 52% complete after 5 h (eq 12). The reaction of 3a with 4a catalyzed by 2 mol % of the orthoOMe catalyst 2b was 85% complete after 4 h at room temperature, while the same reaction conducted at 50 °C was 77% complete after 4 h (eq 13). Because these reactions proceeded in the absence of added K3PO4, we conclude that the methyl carbonate or methoxide generated in situ can serve as the base to form the benzimidazolate nucleophile and that base-free allylations of heterocyclic nucleophiles can be developed. Because higher conversions were observed for reactions conducted in the absence of K3PO4 at room temperature than from those at 50 °C, we conclude that increasing catalyst deactivation occurs during the reactions with increasing temperature. Second, competition experiments that are analogous to those shown in Scheme 6, but conducted in the absence of K3PO4, generate product ratios that contain more of the species derived from the less acidic nitrogen nucleophile.82 This observation, along with the observation of reaction in the absence of exogenous base suggest that both the methyl carbonate (or methoxide) generated in situ and K3PO4 contribute to the formation of the azolate nucleophile. Although we are cognizant of the limited solubility of K3PO4 in THF solvent and the close proximity of the counterions to the allyliridium intermediate, we suggest that reactions conducted with added K3PO4 contain a higher concentration of azolate anion than reactions conducted without added K3PO4. An increase in the concentration of azolate anion would increase the rate of the N-allylation of (82) Product ratios for competition experiments in the absence of potassium phosphate base are as follows: 18:82 (11a/28) for the competition reaction between imidazole and benzyl amine; 40:60 (11a/29) for the competition reaction between imidazole and aniline; 50:50 (11a/5a) for the competition reaction between imidazole and benzimidazole; and 30:70 (11a/21k) for the competition reaction between imidazole and bis-BOC-adenine.

Ir-Catalyzed N-Allylation of Heterocycles

ARTICLES

Figure 5. A comparison of catalysts [Ir(COD)(κ2-L1)(L1)] (1) and

Figure 6. A comparison of catalysts [Ir(COD)(κ2-L1)(ethylene)] (2a) and

[Ir(COD)(κ2-L2)(L2)] (30) for the reaction of methyl cinnamyl carbonate with benzimidazole in the presence and absence of K3PO4. All reactions were conducted in THF solvent at room temperature. Catalysts and conditions: 9, 2 mol % 1 + 1 mol % [Ir(COD)Cl]2 (no base); 2, 2 mol % 30 + 1 mol % [Ir(COD)Cl]2 (no base); b, 2 mol % 1 + [Ir(COD)Cl]2 + 1.0 equiv of K3PO4; [, 2 mol % 30 + 1 mol % [Ir(COD)Cl]2 + 1.0 equiv of K3PO4.

[Ir(COD)(κ2-L2)(ethylene)] (2b) for the reaction of methyl cinnamyl carbonate with benzimidazole in the presence and absence of K3PO4. All reactions were conducted in THF solvent at room temperature. Catalysts and conditions: 9, 2 mol % 2a (no base); 2, 2 mol % 2b (no base); b, 2 mol % 2a + 1.0 equiv of K3PO4; [, 2 mol % 2b + 1.0 equiv of K3PO4.

nucleophiles containing more acidic N-H bonds, and this prediction is consistent with the larger amounts of product derived from the more acidic nitrogen nucleophiles in reactions conducted with added K3PO4.

Scheme 8. Cyclometallation to Form Iridium Metallacycles 2a and 2b and Ring Opening in the Presence of a Protic Acid

Studies of the Relative Activity and Stability of Catalysts 2a and 2b. Kinetic studies of the reactions of benzimidazole

with methyl cinnamyl carbonate catalyzed by bis-phosphoramidite complexes [Ir(COD)(κ2-L1)(L1)] (1) and [Ir(COD)(κ2L2)(L2)] (30) were conducted to gauge the effect of substituents on the phosphoramidite ligand on catalyst stability and activity in the presence and absence of K3PO4. A comparison of reactions in the presence and absence of K3PO4 catalyzed by the combination of [Ir(COD)Cl]2 (to bind the κ1-phosphoramidite after dissociation) and complex 1 or 30 is shown in Figure 5. As previously observed during the development of conditions for the reaction of benzimidazole with methyl cinnamyl carbonate (see Table 1, entry 1), the reactions catalyzed by bisphosphoramidite complexes 1 and 30 occurred to low conversions in the absence of K3PO4. However, analogous reactions at room temperature occurred to high conversions when conducted in the presence of 1.0 equiv of K3PO4. These data show that deactivation of either the metallacyclic catalysts 1 and 30 or the [Ir(COD)Cl]2 occurred rapidly in the absence of exogenous base. Furthermore, these data suggest that substituents on the arylethyl group of the phosphoramidite ligand have minimal impact on the stability and activity of metallacyclic bis-phosphoramidite catalysts 1 and 30. In contrast to the reactions catalyzed by bis-phosphoramidite complexes 1 and 30, the reactions catalyzed by metallacyclic ethylene adducts [Ir(COD)(κ2-L1)(ethylene)] (2a) and [Ir(COD)(κ2-L2)(ethylene)] (2b) (Figure 6) were significantly affected by the substituents on the arylethyl group of the phosphoramidite ligand. The reaction of methyl cinnamyl carbonate with benzimidazole in the presence of parent catalyst 2a without added K3PO4 occurred to modest conversion, but the same reaction catalyzed by ortho-OMe catalyst 2b occurred to nearly complete conversion. This result implies that the metallacyclic iridium complex 2b generated from the orthoanisyl phosphoramidite ligand L2 is more stable toward decomposition than the metallacyclic iridium complex 2a generated from the parent phosphoramidite L1. In contrast, the kinetic profile of the reactions catalyzed by complexes 2a and 2b are similar to each other in the presence of added K3PO4. This result implies that the phosphate base prevents decomposi-

tion of the catalyst and that the rates of the sets of reactions within the catalytic cycle involving the species generated from the ethylene adducts 2a and 2b are similar to each other. The latter conclusion contrasts a previous claim that the ortho-anisyl phosphoramidite ligand L2 generates a more active metallacyclic iridium catalyst than does phosphoramidite ligand L1.32 We propose that the metallacyclic catalysts undergo decomposition by protonation of the metallacyclic unit to form an inactive acyclic species. Metallacyclic iridium complexes 2a and 2b are formed from the four-coordinate iridium complexes 31a and 31b and ethylene in the presence of an amine base, and opening of the metallacycle occurs upon addition of a protic acid, such as acetic acid or excess amine hydrochloride (Scheme 8).17 Thus, metallacyclic iridium catalysts 2a and 2b would be expected to be stable during allylic substitution reactions of basic nitrogen nucleophiles but to undergo ring opening as a side reaction during allylic substitution reactions of the acidic nitrogen nucleophiles studied in this work. To assess this proposal for the mechanism of decomposition of the iridium metallacycles in the presence of the imidazole nucleophiles, we studied the stoichiometric reactions of catalysts 2a and 2b with excess benzimidazole (pKa ) 16.4, DMSO) and bis-BOC-adenine (pKa ≈ 15, DMSO). Consistent with our conclusion about the relative stabilities of catalysts 2a and 2b, parent catalyst 2a reacted with the two azoles more rapidly than did the ortho-OMe catalyst 2b. The half-life of catalyst 2a in THF at 50 °C in the presence of 25 equiv of benzimidazole was ∼60 min, while the half-life of catalyst 2b was greater than 4 h under identical conditions. Furthermore, the half-life of catalyst 2a in the presence of 25 equiv of bis-BOC adenine was less than 20 min, while the half-life of catalyst 2b was ∼40 min. Analogous reactions of 2a and 2b with N-methyl benzimidazole and 9-methyl-bis-BOC-adenine resulted in minimal decomposition of 2a or 2b. J. AM. CHEM. SOC.

9

VOL. 131, NO. 25, 2009

8981

ARTICLES

Stanley and Hartwig

Scheme 9. Mechanism for the Deactivation of [Ir(COD)(κ2-L1)(Ethylene)] (2a) in the Presence of Benzimidazole

Scheme 10. Independent Generation of [Ir(COD)(L1)(benzimidazolate)] 32 and Its Rapid Decomposition to Free Phosphoramidite Ligand L1 and [Ir(COD)(benzimidazolate)]3

The stoichiometric reaction of parent catalyst 2a with excess benzimidazole was studied in further depth to develop a more detailed understanding of the reactions leading to catalyst decomposition (Scheme 9). Three major new species were observed by 31P{1H} NMR spectroscopy from this reaction. Two sets of doublets that correspond to [Ir(COD)(κ2-L1)(L1)] (1) were observed at 153.5 ppm (J ) 47 Hz) and 128.5 ppm (J ) 47 Hz), and a singlet corresponding to the free phosphoramidite ligand L1 was observed at 151.2 ppm. In addition, a singlet at 120.0 ppm, which we propose to correspond to [Ir(COD)(L1)(benzimidazolate)] (32), was observed transiently. Attempts to independently synthesize complex 32 from the reaction of [Ir(COD)(L1)Cl] (31a) with sodium benzimidazolate led to rapid formation of free phosphoramidite ligand L1 and the known complex [Ir(COD)(benzimidazolate)]3 as a yellow precipitate (Scheme 10).83 Based on these data, we propose that catalyst 2a reacts with benzimidazole to form benzimidazolate complex 32 as a transient intermediate, either by direct protonation of the metallacycle or by oxidative addition of the azole N-H bond,84 followed by reductive elimination to form a C-H bond. Complex 32 decomposes to form free phosphormidite L1 and [Ir(COD)(benzimidazolate)]3. The free phosphoramidite ligand (83) Ramaswamy, Y. S.; Halesha, R.; Gowda, N. M. N.; Reddy, G. K. N. Indian J. Chem., Sec. A 1991, 30A, 393–399. (84) Cuenca, T.; Padilla, A.; Royo, P.; Parra-Hake, M.; Pellinghelli, M. A.; Tiripicchio, A. Organometallics 1995, 14, 848–854. 8982

J. AM. CHEM. SOC.

9

VOL. 131, NO. 25, 2009

Table 7. Base-Free N-Allylation of Benzimidazole and Imidazole Nucleophilesa

entry

azole (R)

time (h)

product

yield (%)b

5/6 or 12/13c

ee (%)d

1 2 3 4 5

4a (H) 4b (Ph) 4d (Cl) 11a (H) 11b (Ph)

7 7 4 20 5

5a 5j 5m 12a 12b

87 87 84 92 86

97:3 96:4 92:8 99:1 95:5

98 98 97 98 99

a See Supporting Information for experimental details. b Isolated yield of 5 or 12. c Determined by 1H NMR spectroscopy of the crude reaction mixture. d Determined by chiral HPLC.

L1 then reacts with ethylene adduct 2a to form [Ir(COD)(κ2L1)(L1)] (1), which is known to catalyze the allylic substitution reaction with slow rates in the absence of an additive to sequester the κ1-bound phosphoramidite ligand.38 The ortho-OMe catalyst 2b is more stable toward benzimidazole than the parent catalyst 2a. After 4 h at 50 °C, the only decomposition product (ca. 30%) observed by 31P{1H} NMR spectroscopy from the stoichiometric reaction of 2b with excess benzimidazole was the free phosphoramidite ligand L2 at 155.4 ppm. [Ir(COD)(κ2-L2)(L2)] (30), the ortho-anisyl analogue of [Ir(COD)(κ2-L1)(L1)] (1), was not observed in the reaction of 2b with benzimidazole. This result implies that the increased stability of ortho-OMe catalyst 2b, relative to the parent catalyst 2a, results from two factors. First, the rate of ring opening of the metallacycle in ortho-OMe catalyst 2b is slower than that for parent catalyst 2a. Second, ortho-OMe catalyst 2b does not readily react with free phosphoramidite ligand L2 to form the less active catalyst [Ir(COD)(κ2-L2)(L2)] (30). These factors result in a concentration of 2b that is greater than that of 2a in catalytic reactions of azole nucleophiles with allylic carbonates in the absence of K3PO4. Furthermore, these results are consistent with rates of reactions of benzimidazole with methyl cinnamyl carbonate in the absence of K3PO4 base that are faster when catalyzed by ortho-OMe complex 2b than when catalyzed by parent complex 2a (Figure 6). Base-Free N-Allylation of Benzimidazole and Imidazole Nucleophiles. The stability of catalyst 2b toward deactivation

in the presence of benzimidazole led us to study reactions of benzimidazole and imidazole nucleophiles in the absence of added K3PO4 (Table 7). The room temperature reactions of methyl cinnamyl carbonate with benzimidazole, 2-phenylbenzimidazole, and 2-chlorobenzimidazole in the presence of 2 mol % of the ortho-OMe catalyst 2b occurred to high conversions and formed products 5a, 5j, and 5m in high yields (>84%) with high selectivities (branched-to-linear >92:8 and >97% ee) (entries 1-3). Imidazole nucleophiles, which contain a less acidic N-H bond than the benzimidazoles, also reacted selectively with 3a (entries 4 and 5). Products 12a and 12b were formed in greater than 86% yield with >95:5 branched-to-linear selectivity and >98% ee.

Ir-Catalyzed N-Allylation of Heterocycles Scheme 11. Base-Free N-Allylation of Bis-BOC-adenine 20d with

3a

The instability of catalyst 2b to bis-BOC adenine in the absence of K3PO4 implied that base-free N-allylations of purine nucleophiles would not occur to the same high conversions as were observed for reactions of the benzimidazoles and imdazoles. Indeed, the reaction of bis-Boc-adenine with methyl cinnamyl carbonate occurred to only 63% conversion, and product 21k was isolated in only 49% yield (Scheme 11). Thus, base-free, iridium-catalyzed N-allylations of benzimidazole and imidazole nucleophiles with allylic carbonates occur in high yield and selectivity with catalyst 2b, but allylations of purine nucleophiles should be conducted with added K3PO4 to prevent deactivation of metallacyclic iridium catalysts 2a and 2b. Conclusions

In summary, we have developed the first catalytic enantioselective allylation of benzimidazole, imidazole, and purine nucleophiles with achiral acyclic allylic electrophiles to form chiral N-allylated azoles. These reactions are enabled by the use of single-component ethylene catalysts 2a and 2b. Reactions with these single-component catalysts occur with less isomerization and with less iridium than those with previous metallacyclic iridium catalysts. The N-allylated benzimidazoles and purines generated from these reactions are readily converted to substructures, such as R-benzimidazolyl alcohols, β-benzimidazolyl alcohols, β-benzimidazolyl acids, R-purinyl alcohols, and β-purinyl alcohols, present in a variety of biologically important compounds. Furthermore, a formal synthesis of a JNK3 inhibitor demonstrates that iridium-catalyzed allylation provides rapid access to N-allylated imidazoles that can be transformed into enantiomerically enriched annulated products by intramolecular addition of an imidazole C(2)-H bond across the pendant alkene. In addition, base-free N-allylation of benzimidazole and imidazole nucleophiles occurs readily in the presence of 2b as catalyst. Competition experiments between azole nucleophiles and common amine nucleophiles helped elucidate the factors that

ARTICLES

lead to the ability to conduct the N-allylation of azole nucleophiles, even though they are much weaker nucleophiles than benzylamines, alkylamines, and arylamines in stoichiometric reactions with organic electrophiles. More specifically, the ratio of rates from competition reactions of amines versus azoles is much smaller than that for reaction of the two types of neutral nucleophiles toward organic electrophiles. Because this ratio of rates depended on the presence or absence of added base, we suggest that the deprotonated forms of imidazole, benzimidazole, and adenine nucleophiles react, at least in part, as the nucleophile in the iridium-catalyzed N-allylation reactions. This hypothesis also rationalizes the preferential reaction of the more acidic azoles in competition studies conducted in the presence of base. The same trend is not observed when comparing the rates of separate allylation reactions of the different classes of azoles because the more acidic heterocycles led to faster catalyst decomposition. Kinetic studies of reactions catalyzed by metallacyclic complexes 2a and 2b show that the rates of reactions catalyzed by the parent catalyst 2a and the ortho-OMe-substituted catalyst 2b are similar but that catalyst 2b is more stable to acidic benzimidazole, imidazole, and purine nucleophiles than catalyst 2a. Stoichiometric reactions of the metallacyclic complexes 2a and 2b with azoles show that ring opening of the metallacycles occurs in the presence of the azoles to form [Ir(COD)(azolate)]n and free phosphoramidite ligand L1 or L2 and that ring opening of the metallacycle in the ortho-OMe-substituted complex 2b is slower than that of the metallacycle in parent complex 2a. In addition, complex 2a reacts with free ligand L1 to generate the less active catalyst [Ir(COD)(κ2-L1)(L1)] (1). This process is not observed in stoichiometric reactions of complex 2b with azole nucleophiles, and the absence of this process provides further evidence of the improved stability of ortho-OMesubstituted catalyst 2b, relative to that of parent catalyst 2a. Efforts to develop more robust iridium catalysts that will extend the method to additional azole nucleophiles and other acidic nitrogen nucleophiles are currently underway. Acknowledgment. We thank the NIH for financial support of this work (NIH GM55382 to J.F.H. and GM84584 to L.M.S.) and Johnson-Matthey for gifts of [Ir(COD)Cl]2 and IrCl3. We thank Dr. Klaus Ditrich and BASF for a gift of both enantiomers of 1-(2methoxyphenyl)ethylamine. We thank Mark Pouy and Dr. Dan Weix for experimental assistance and insightful discussions. Supporting Information Available: Experimental procedures, characterization data, and complete ref 48. This material is available free of charge via the Internet at http://pubs.acs.org. JA902243S

J. AM. CHEM. SOC.

9

VOL. 131, NO. 25, 2009

8983